arXiv subjects

Zihao Yang

Publications and source records attributed to Zihao Yang.

At least 19 recordsLinked to original sources

arXiv2026

InfluenceField: A Differentiable Field with Interventionally Identifiable Causal Structure for Multimodal World Modeling

Multimodal large language models often capture visual-linguistic correlations but struggle to predict how local visual interventions propagate and affect downstream answers. We introduce InfluenceField, an intervention-aware latent field inserted between the visual encoder and language decoder. It lifts patch features into a continuous spatial representation, propagates directed influence over multiple steps, and predicts local intervention effects through a shared transition operator. Training jointly optimizes language modeling, cross-environment invariance, counterfactual rollout supervision, and structural regularization. For a nonlinear finite-basis population model, we show that target-aligned interventional supervision, together with a one-step separation condition on the transition, restricts admissible representations to within-location reparameterizations, so that the directed dependency graph of the full transition is recovered exactly. A linear specialization gives an exact partial-coverage characterization and a finite-loss stability bound, and the field analysis derives the spatial profile of coefficient interventions together with a shared-channel calibration result. On CausalVQA, InfluenceField improves overall accuracy over its backbone by 13.1 percentage points, with the largest gains on the planning and hypothetical categories. Capacity-matched baselines and structural controls attribute the gains in robustness and factual-counterfactual consistency to the causal objectives rather than to added capacity.

cs.LG↗

arXiv2026

Bioinfoysis Technical Report

Large language model agents have shown promise in bioinformatics, but most existing systems focus primarily on producing final answers, treating planning, tool use, and code execution as transient interactions. This design is poorly suited to long-horizon bioinformatics tasks, where conclusions must remain connected to the data, computations, and intermediate evidence that support them. We introduce \textbf{Bioinfoysis}, a multi-agent harness that represents each request as a persistent, artifact-grounded analysis run. Bioinfoysis combines global planning with step-wise, evidence-driven replanning: the planner maintains an executable checklist and revises pending steps using structured handoffs returned after each worker execution. These handoffs bind intermediate results to their responsible agent, checklist step, and plan generation, preventing stale evidence from being silently reused after replanning. A controlled runtime validates generated scripts, tables, and figures before they are used in downstream analysis or reporting, while role-specific context, persistent memory, and governed bioinformatics skills support reliable execution over long analysis trajectories. We evaluate Bioinfoysis on BixBench and two question-answering tracks of LAB-Bench 2. On BixBench, Bioinfoysis achieves state-of-the-art accuracy of 82.4\%. Across four underlying language models, Bioinfoysis increases average accuracy from 27.81\% to 64.13\% on SeqQA2 and from 3.13\% to 31.25\% on DbQA2. These results demonstrate that reliable bioinformatics automation depends not only on model capability, but also on the harness that governs planning, execution, memory, and evidence flow. We hope that the emergence of Bioinfoysis will play a driving and leading role in the development of the bioinformatics community. Our demo website can be seen in https://report.bioinfoysis.com/.

cs.AI↗

arXiv2026

TemporalGS: Training-Free Plug-and-Play Acceleration for 3D Gaussian Splatting Rendering via Temporal Priors

3D Gaussian Splatting (3DGS) has revolutionized novel-view synthesis with its fast and high-fidelity rendering. However, rendering at high FPS and low latency across various scenes remains a challenge, especially when large amounts of 3D Gaussian ellipsoids appear in the scene. To address this issue, we introduce TemporalGS, to the best of our knowledge, the first training-free plug-and-play algorithmic approach to accelerate 3DGS rendering without any post-training or post-processing, implemented on top of tile-based software rasterization. The key idea is that, instead of rendering frames independently as 3DGS, we leverage the temporal priors, represented by novel geometry and appearance buffers, etc., to reduce redundancy of Gaussian preprocessing, sorting, and rasterization operations of consecutive frames. Specifically, we propose two acceleration strategies: (1) temporal dynamic culling, which filters out Gaussians that contribute less to current frame rendering; (2) selective rendering, which renders only a small portion of tiles that cannot be approximated by the temporal priors. By adapting and interleaving these two strategies, TemporalGS yields a simple but effective plug-and-play solution for 3DGS rendering speed-up without any training. Extensive experiments show that TemporalGS achieves comparable or even better performance compared to existing state-of-the-art post-training or post-processing-based 3DGS rendering acceleration approaches. TemporalGS can significantly enhance the rendering speed of various 3DGS methods, achieving up to $1.48\times$ acceleration, while maintaining competitive rendering quality. We further extend our TemporalGS to hardware rasterization-based 3DGS to show the portability of our algorithm.

cs.CV↗

arXiv2026

From Item-Only to Query-Item: Query-Conditioned Generative Search with QGS in Quark

Generative sequence models have shown strong results in recommendation. Applying them to search ranking is more challenging. Search behavior is inherently query-driven. Each query switch introduces a sharp topic shift in the user's interaction history. Existing generative methods flatten queries and items into a single token sequence. They do not distinguish query boundaries. This causes the model to mix different query intents into one prediction target, resulting in noisy supervision. We present Query-Conditioned Generative Search (QGS). QGS encodes each interaction as a (query, item) pair token. It trains with a query-conditioned next-item objective. The prediction target changes from a noisy marginal P(item_{t+1}|context_{<=t}) to a clean conditional P(item_{t+1}|context_{<=t}, query_{t+1}). This directly removes the semantic discontinuity caused by query switches. Encoding long interaction histories with standard attention has quadratic cost. This is impractical under strict online latency budgets. We introduce a Linear HSTU encoder. It replaces full attention with causal linear recurrence. Per-layer complexity drops from O(L^2) to O(L) with no loss in ranking quality. Traditional search ranking depends on hand-crafted features like text-matching scores, statistical signals, and behavioral features. We propose HFG-Attention to preserve them in the generative framework. It organizes heterogeneous features into semantic groups and fuses them through a dedicated attention block. This bridges sparse engineered signals with dense sequential representations. QGS is deployed in the ranking module of Quark Search, a major commercial search engine in China. Online A/B tests show statistically significant gains: +0.62% CTR, +0.38% Click-Search Ratio, and +3.55% PV Duration over the production deep learning baseline.

cs.IR↗

arXiv2026

HorizonDrive: Self-Corrective Autoregressive World Model for Long-horizon Driving Simulation

Closed-loop driving simulation requires real-time interaction beyond short offline clips, pushing current driving world models toward autoregressive (AR) rollout. Existing AR distillation approaches typically rely on frame sinks or student-side degradation training. The former transfers poorly to driving due to fast ego-motion and rapid scene changes, while the latter remains bounded by the teacher's single-pass output length and thus provides only a limited supervision horizon. A natural question is: can the teacher itself be extended via AR rollout to provide unbounded-horizon supervision at bounded memory cost? The key difficulty is that a standard teacher drifts under its own predictions, contaminating the supervision it provides. Our key insight is to make the teacher rollout-capable, ensuring reliable supervision from its own AR rollouts. This is instantiated as HorizonDrive, an anti-drifting training-and-distillation framework for AR driving simulation. First, scheduled rollout recovery (SRR) trains the base model to reconstruct ground-truth future clips from prediction-corrupted histories, yielding a teacher that remains stable across long AR rollouts. Second, the rollout-capable teacher is extended via AR rollout, providing long-horizon distribution-matching supervision under bounded memory, while a short-window student aligns to it with teacher rollout DMD (TRD) for efficient real-time deployment. HorizonDrive natively supports minute-scale AR rollout under bounded memory; on nuScenes, HorizonDrive reduces FID by 52% and FVD by 37%, and lowers ARE and DTW by 21% and 9% relative to the strongest long-horizon streaming baselines, while remaining competitive with single-pass driving video generators.

cs.CV↗

arXiv2026

Compared to What? Baselines and Metrics for Counterfactual Prompting

Counterfactual prompting (i.e., perturbing a single factor and measuring output change) is widely used to evaluate things like LLM bias and CoT faithfulness. But in this work we argue that observed effects cannot be attributed to the targeted factor without accounting for baseline "meaning-preserving" modifications to text that establish general model sensitivity. This is because every counterfactual edit is a compound treatment that bundles the variable of interest with incidental surface-form variation; this violates treatment variation irrelevance. We observe prediction flip rates on MedQA of 14.9% when we surgically change patient gender. However, this is statistically indistinguishable from the flip rates induced by simply paraphrasing inputs (14.1%). In this case, it would therefore be unwarranted to conclude that the LLM is especially sensitive to patient gender. To account for this and robustly measure the effects of targeted interventions, we propose a framework in which we compare (via statistical testing) differences observed under target interventions to those induced by paraphrasing inputs. We then use this framework to revisit a analysis done on the MedPerturb dataset, which reported evidence of model sensitivity to patient demographics and stylistic cues. We find that these effects largely dissipate when we account for general model sensitivity, with only 5 of 120 tests reaching statistical significance. Applying the same framework to occupational biography classification, we detect clearly significant directional gender bias, showing that the framework identifies real directional effects even when they are small. We evaluate a range of metrics (aggregate, per-sample distributional, and regression) and find that per-sample metrics are dramatically more powerful than aggregate metrics and regression powerfully and uniquely characterizes effect direction and magnitude.

cs.CL↗

arXiv2026

Driving with DINO: Vision Foundation Features as a Unified Bridge for Sim-to-Real Generation in Autonomous Driving

Driven by the emergence of Controllable Video Diffusion, existing Sim2Real methods for autonomous driving video generation typically rely on explicit intermediate representations to bridge the domain gap. However, these modalities face a fundamental Consistency-Realism Dilemma. Low-level signals (e.g., edges, blurred images) ensure precise control but compromise realism by "baking in" synthetic artifacts, whereas high-level priors (e.g., depth, semantics, HDMaps) facilitate photorealism but lack the structural detail required for consistent guidance. In this work, we present Driving with DINO (DwD), a novel framework that leverages Vision Foundation Module (VFM) features as a unified bridge between the simulation and real-world domains. We first identify that these features encode a spectrum of information, from high-level semantics to fine-grained structure. To effectively utilize this, we employ Principal Subspace Projection to discard the high-frequency elements responsible for "texture baking," while concurrently introducing Random Channel Tail Drop to mitigate the structural loss inherent in rigid dimensionality reduction, thereby reconciling realism with control consistency. Furthermore, to fully leverage DINOv3's high-resolution capabilities for enhancing control precision, we introduce a learnable Spatial Alignment Module that adapts these high-resolution features to the diffusion backbone. Finally, we propose a Causal Temporal Aggregator employing causal convolutions to explicitly preserve historical motion context when integrating frame-wise DINO features, which effectively mitigates motion blur and guarantees temporal stability. Project page: https://albertchen98.github.io/DwD-project/

cs.CV↗

arXiv2026

Measuring the Coronal Magnetic Field with 2D Coronal Seismology: A Forward-Modeling Validation

In recent years, a two-dimensional (2D) coronal seismology technique applied to spectral-imaging data from the Coronal Multi-channel Polarimeter (CoMP) and UCoMP has enabled routine measurement of the global coronal magnetic field. The technique combines coronal transverse wave phase speed from Doppler measurements with electron densities from the Fe \sc{xiii}\rm{} 10798/10747 \AA\ intensity ratio to infer the magnetic field strength, while the wave propagation directions from Doppler measurements trace the magnetic field direction. To validate the accuracy and robustness of this method, we use forward modeling of a MURaM simulation that produces open and closed magnetic structures with excited waves. From the synthetic Doppler velocity, Fe \sc{xiii}\rm{} infrared line intensities, and linear polarization signals, we apply the 2D coronal seismology technique to estimate the magnetic field strength and direction. A comparison with the simulation ground truth shows close agreement, indicating that the technique can recover the line-of-sight emissivity-weighted magnetic field direction and strength with high accuracy. We also perform a parameter-space analysis to quantify sensitivities of the method to parameter choice. These findings provide practical guidance for CoMP/UCoMP-like analysis and demonstrate that 2D coronal seismology can deliver reliable, LOS emissivity-weighted measurements of the coronal magnetic field from coronal wave observations.

astro-ph.SR↗

arXiv2025

Efficient LLM Safety Evaluation through Multi-Agent Debate

Safety evaluation of large language models (LLMs) increasingly relies on LLM-as-a-judge pipelines, but strong judges can still be expensive to use at scale. We study whether structured multi-agent debate can improve judge reliability while keeping backbone size and cost modest. To do so, we introduce HAJailBench, a human-annotated jailbreak benchmark with 11,100 labeled interactions spanning diverse attack methods and target models, and we pair it with a Multi-Agent Judge framework in which critic, defender, and judge agents debate under a shared safety rubric. On HAJailBench, the framework improves over matched small-model prompt baselines and prior multi-agent judges, while remaining more economical than GPT-4o under the evaluated pricing snapshot. Ablation results further show that a small number of debate rounds is sufficient to capture most of the gain. Together, these results support structured, value-aligned debate as a practical design for scalable LLM safety evaluation.

cs.AI↗

arXiv2025

Cambrian-S: Towards Spatial Supersensing in Video

We argue that progress in true multimodal intelligence calls for a shift from reactive, task-driven systems and brute-force long context towards a broader paradigm of supersensing. We frame spatial supersensing as four stages beyond linguistic-only understanding: semantic perception (naming what is seen), streaming event cognition (maintaining memory across continuous experiences), implicit 3D spatial cognition (inferring the world behind pixels), and predictive world modeling (creating internal models that filter and organize information). Current benchmarks largely test only the early stages, offering narrow coverage of spatial cognition and rarely challenging models in ways that require true world modeling. To drive progress in spatial supersensing, we present VSI-SUPER, a two-part benchmark: VSR (long-horizon visual spatial recall) and VSC (continual visual spatial counting). These tasks require arbitrarily long video inputs yet are resistant to brute-force context expansion. We then test data scaling limits by curating VSI-590K and training Cambrian-S, achieving +30% absolute improvement on VSI-Bench without sacrificing general capabilities. Yet performance on VSI-SUPER remains limited, indicating that scale alone is insufficient for spatial supersensing. We propose predictive sensing as a path forward, presenting a proof-of-concept in which a self-supervised next-latent-frame predictor leverages surprise (prediction error) to drive memory and event segmentation. On VSI-SUPER, this approach substantially outperforms leading proprietary baselines, showing that spatial supersensing requires models that not only see but also anticipate, select, and organize experience.

cs.CV↗

arXiv2025

iPanda: An LLM-based Agent for Automated Conformance Testing of Communication Protocols

Conformance testing is essential for ensuring that protocol implementations comply with their specifications. However, traditional testing approaches involve manually creating numerous test cases and scripts, making the process labor-intensive and inefficient. Recently, Large Language Models (LLMs) have demonstrated impressive text comprehension and code generation abilities, providing promising opportunities for automation. In this paper, we propose iPanda, the first framework that leverages LLMs to automate protocol conformance testing. Given a protocol specification document and its implementation, iPanda first employs a keyword-based method to automatically generate comprehensive test cases. Then, it utilizes retrieval-augmented generation and customized CoT strategy to effectively interpret the implementation and produce executable test programs. To further enhance programs' quality, iPanda incorporates an iterative optimization mechanism to refine generated test scripts interactively. Finally, by executing and analyzing the generated tests, iPanda systematically verifies compliance between implementations and protocol specifications. Comprehensive experiments on various protocols show that iPanda significantly outperforms pure LLM-based approaches, improving the success rate (Pass@1) of test-program generation by factors ranging from 4.675 times to 10.751 times.

cs.SE↗

arXiv2025

Beyond Winning: Margin of Victory Relative to Expectation Unlocks Accurate Skill Ratings

Knowledge of accurate relative skills in any competitive system is essential, but foundational approaches such as ELO discard extremely relevant performance data by concentrating exclusively on binary outcomes. While margin of victory (MOV) extensions exist, they often lack a definitive method for incorporating this information. We introduce Margin of Victory Differential Analysis (MOVDA), a framework that enhances traditional rating systems by using the deviation between the true MOV and a $\textit{modeled expectation}$. MOVDA learns a domain-specific, non-linear function (a scaled hyperbolic tangent that captures saturation effects and home advantage) to predict expected MOV based on rating differentials. Crucially, the $\textit{difference}$ between the true and expected MOV provides a subtle and weighted signal for rating updates, highlighting informative deviations in all levels of contests. Extensive experiments on professional NBA basketball data (from 2013 to 2023, with 13,619 games) show that MOVDA significantly outperforms standard ELO and Bayesian baselines. MOVDA reduces Brier score prediction error by $1.54\%$ compared to TrueSkill, increases outcome accuracy by $0.58\%$, and most importantly accelerates rating convergence by $13.5\%$, while maintaining the computational efficiency of the original ELO updates. MOVDA offers a theoretically motivated, empirically superior, and computationally lean approach to integrating performance magnitude into skill rating for competitive environments like the NBA.

stat.ML↗

arXiv2025

Gravitational wave signatures of preheating in Higgs--$R^2$ inflation

We present a comprehensive analysis of the preheating dynamics and associated gravitational wave signatures in the Higgs--$R^2$ inflationary model. Using lattice simulations, we investigate the post-inflationary evolution of the system across the parameter space, covering both the Higgs-like and $R^2$-like scenarios. We demonstrate that the efficiency of preheating is significantly dependent on the nonminimal coupling parameter $\xi$. As the $\xi$ parameter increases, moving from the $R^2$-like regime to the Higgs-like regime, we observe more efficient preheating. Through detailed numerical computations, efficient preheating is shown to lead to stronger gravitational wave production. The amplitude of the gravitational wave spectrum varies by several orders of magnitude as we move from the $R^2$-like regime to the Higgs-like regime. The resultant gravitational wave signatures can serve as a potential observational probe to distinguish between different parameter regimes of the Higgs--$R^2$ model.

astro-ph.CO↗

arXiv2025

Random batch list method for metallic system with embedded atom potential

The embedded atom method (EAM) is one of the most widely used many-body, short-range potentials in molecular dynamics simulations, particularly for metallic systems. To enhance the efficiency of calculating these short-range interactions, we extend the random batch list (RBL) concept to the EAM potential, resulting in the RBL-EAM algorithm. The newly presented method introduces two "core-shell" lists for approximately computing the host electron densities and the force terms, respectively. Direct interactions are computed in the core regions, while in the shell zones a random batch list is used to reduce the number of interaction pairs, leading to significant reductions in both computational complexity and storage requirements. We provide a theoretical, unbiased estimate of the host electron densities and the force terms. Since metallic systems are Newton-pair systems, we extend the RBL-EAM algorithm to exploit this property, thereby halving the computational cost. Numerical examples, including the lattice constant, the radial distribution function, and the elastic constants, demonstrate that the RBL-EAM method significantly accelerates simulations several times without compromising accuracy.

cond-mat.mtrl-sci↗

arXiv2025

Scaffold-Assisted Window Junctions for Superconducting Qubit Fabrication

The superconducting qubit is one of the promising directions in realizing fault-tolerant quantum computing (FTQC), which requires many high-quality qubits. To achieve this, it is desirable to leverage modern semiconductor industry technology to ensure quality, uniformity, and reproducibility. However, conventional Josephson junction fabrication relies mainly on resist-assistant double-angle evaporation, posing integration challenges. Here, we demonstrate a lift-off-free qubit fabrication that integrates seamlessly with existing industrial technologies. This method employs a silicon oxide (SiO$_2$) scaffold to define an etched window with a well-controlled size to form a Josephson junction. The SiO$_2$, which has a large dielectric loss, is etched away in the final step using vapor HF leaving little residue. This Window junction (WJ) process mitigates the degradation of qubit quality during fabrication and allows clean removal of the scaffold. The WJ process is validated by inspection and Josephson junction measurement. The scaffold removal process is verified by measuring the quality factor of the resonators. Furthermore, compared to scaffolds fabricated by plasma-enhanced chemical vapor deposition (PECVD), qubits made by WJ through physical vapor deposition (PVD) achieve relaxation time up to $57\,\mu\text{s}$. Our results pave the way for a lift-off-free qubit fabrication process, designed to be compatible with modern foundry tools and capable of minimizing damage to the substrate and material surfaces.

physics.app-ph↗

arXiv2025

BPQA Dataset: Evaluating How Well Language Models Leverage Blood Pressures to Answer Biomedical Questions

Clinical measurements such as blood pressures and respiration rates are critical in diagnosing and monitoring patient outcomes. It is an important component of biomedical data, which can be used to train transformer-based language models (LMs) for improving healthcare delivery. It is, however, unclear whether LMs can effectively interpret and use clinical measurements. We investigate two questions: First, can LMs effectively leverage clinical measurements to answer related medical questions? Second, how to enhance an LM's performance on medical question-answering (QA) tasks that involve measurements? We performed a case study on blood pressure readings (BPs), a vital sign routinely monitored by medical professionals. We evaluated the performance of four LMs: BERT, BioBERT, MedAlpaca, and GPT-3.5, on our newly developed dataset, BPQA (Blood Pressure Question Answering). BPQA contains $100$ medical QA pairs that were verified by medical students and designed to rely on BPs . We found that GPT-3.5 and MedAlpaca (larger and medium sized LMs) benefit more from the inclusion of BPs than BERT and BioBERT (small sized LMs). Further, augmenting measurements with labels improves the performance of BioBERT and Medalpaca (domain specific LMs), suggesting that retrieval may be useful for improving domain-specific LMs.

cs.CL↗

arXiv2025

Superlubric sliding ferroelectricity

Sliding ferroelectricity may emerge in many van der Waals bilayers/multilayers and the low switching barriers render ultrafast data writing with low energy cost. We note that such barriers are still much higher compared with structural superlubricity, and in this paper we propose a type of superlubric sliding ferroelectricity in homobilayers separated by a different layer that leads to unprecedented low switching barriers due to incommensurate interfaces. For example, the switching barrier of 3R bilayer MoS2 will be respectively reduced by around 2 or 1 order of magnitudes if they are separated by a graphene or BN monolayer, and the required voltage for switching can be about 1 order of magnitude lower. Such superlubric sliding ferroelectricity widely exists in various similar sandwich trilayer systems where the polarizations stem from symmetry breaking in across-layer stacking configurations, and with ultralow barriers of superlubric sliding, their performances for various applications are greatly enhanced compared with homobilayer sliding ferroelectrics.

cond-mat.mtrl-sci↗

arXiv2024

Measurements of the solar coronal magnetic field based on coronal seismology with propagating Alfvenic waves: forward modeling

Recent observations have demonstrated the capability of mapping the solar coronal magnetic field using the technique of coronal seismology based on the ubiquitous propagating Alfvenic/kink waves through imaging spectroscopy. We established a magnetohydrodynamic (MHD) model of a gravitationally stratified open magnetic flux tube, exciting kink waves propagating upwards along the tube. Forward modeling was performed to synthesize the Fe XIII 1074.7 and 1079.8 nm spectral line profiles, which were then used to determine the wave phase speed, plasma density, and magnetic field with seismology method. A comparison between the seismologically inferred results and the corresponding input values verifies the reliability of the seismology method. In addition, we also identified some factors that could lead to errors during magnetic field measurements. Our results may serve as a valuable reference for current and future coronal magnetic field measurements based on observations of propagating kink waves.

astro-ph.SR↗